OR WAIT 15 SECS
A staged approach to limits should embrace future capabilities.
This paper discusses an approach for the establishment and lifecycle management of biological and biotechnology-derived product specifications. The views presented are consistent with the concept of Quality by Design (QbD), in which critical quality attributes (CQAs) are distinguished from parameters used to monitor process consistency. Specifications and the corresponding limits as applied to CQAs serve to ensure that the product is fit for use, whereas control limits are a manufacturer's tool to monitor shifts and trends in the manufacturing process. In the current paradigm, inappropriate use of specifications creates a disincentive for continuous process understanding; more suitable approaches to analyzing development and manufacturing data are discussed. Statistical methods are presented for deriving and interpreting data against specifications that better manage the risk to the customer of receiving product with diminished safety or efficacy, as well as the risk to the manufacturer of earmarking a satisfactory lot as unacceptable. The recommendations are presented as a rational approach to setting and maintaining specifications, while recognizing that their applicability may not be suitable in all cases, given the heterogeneity of types of regulated biological and biotechnology-derived products and their unique challenges.
The purpose of this paper, which has been developed by the Working Group on Specifications and Formulations of the Pharmaceutical Research and Manufacturers of America (PhRMA) Biologics and Biotechnology Leadership Committee, is to provide guidance on a lifecycle approach to setting global specifications for biological and biotechnology-derived products. In the pharmaceutical industry, specifications are legally binding criteria that a product must meet in order to be marketed. They ensure the consistency and quality of the product and help ensure that it is safe and efficacious over the shelf life of the product. Specifications evolve during product development and ideally should embrace future process capability. This is true for biological and biotechnology-derived products for which there may be limited experience at the time of regulatory filings (including the marketing application), and for which early commercial production often is necessary to gain a better understanding of product quality attributes, methods, and limits.
Currently, there is no industry-wide guideline about the process for establishing specifications for biological and biotechnology-derived products at different stages of the product lifecycle. The International Conference on Harmonization (ICH) Q6B document provides detailed guidance for commercial products and refers to specifications as a list of tests, references to analytical procedures, and appropriate acceptance criteria with numerical limits, ranges, or other criteria to describe the result of the test. Specifications establish a set of criteria to which a drug substance (DS), drug product (DP), or materials at other stages of manufacture should conform to be considered acceptable for use. Conformance to specifications means that the drug substance or drug product, when tested according to the listed analytical procedures, will meet the acceptance criteria. However, this definition may be too restrictive for some applications, such as stability testing or process validation, for which more appropriate means of establishing product quality might be considered.
This paper is laid out in five sections. We will first provide some terminology and definitions. This terminology is not necessarily common throughout the industry, but we hope it will cover all aspects of setting specifications and provide a basis for discussion throughout the paper. The second section of the paper outlines the stages of the lifecycle of a biological or biotechnology-derived product, with emphasis on the level of product information at each stage. The third section describes the components of a product specification, including parameters and components that are unique to this class of products. The fourth section highlights some unresolved issues that must be addressed before setting specifications. The last section proposes a strategy for developing a quality system for biologicals and biotechnology-derived products that helps ensure safety and efficacy to the customer throughout the shelf life of the product, and provides the manufacturer with a powerful set of tools to monitor the manufacturing process.
Sections 1 to 3, covering terminology; the stages of the lifecycle of a product; and components of a biological and biotechnology product specification, appear below, as Part 1 of this article. Section 4 (current issues related to the development of specifications) and section 5 (the suggested approach for developing and maintaining a total quality system), will be published as Parts 2 and 3, in the next two issues of BioPharm International.
A quality attribute is a property that is either demonstrated or predicted to be related to the clinical safety or efficacy of the final product. Among these are purity and potency, which are linked to preclinical and clinical experience during product development. Other properties such as pH and osmolality might be measured to demonstrate the consistency of the manufacturing process.
The reportable value of a quality attribute is the result that is held to appropriate limits. For release testing, this is the value that is reported in the certificate of analysis (COA) for the lot, and may be the average from replicate independent determinations of multiple samples from a lot.
For the purpose of this paper, specification limits will refer to the limits on a quality attribute that predict that the product is fit for use. This is the same as the acceptance criteria set forth in ICH Q6B. Product that does not meet a specification limit for a quality attribute is said to be out-of-specification (OOS).
Control limits, sometimes called process capability limits or alert limits, are limits on a parameter that have been empirically derived from measurements made on product produced in the manufacturing facility, and are used by the manufacturer to monitor a process for shifts or trends during routine manufacture. Product that does not meet its control limits for a parameter is said to be out-of-expectation (OOE) or out-of-trend (OOT). The specification limits for a quality attribute need not be the same as the control limits for that attribute, as the purposes and risks associated with each are different.
It is important to distinguish characterization testing from conformance testing. A significant level of physicochemical product characterization must be performed before establishing specification or control limits. Detailed product characterization may take several years of investigation. Some of these characterization tests will become conformance tests with specification limits, once it has been established that the test measures a property that is related to product quality. Primary among these are tests that may be associated with product safety and efficacy, while others may forecast performance throughout product shelf life. Other characterization tests will be eliminated or reserved for characterization after a process change.
Limits used at release may be different from those used throughout the shelf life of the product. Release limits on a quality attribute are internal or registered limits that forecast that a lot will be fit for use throughout its labeled shelf life. An expiry limit is a limit on the predicted or measured value of the quality attribute beyond which a lot is no longer fit for use.
Preclinical studies are performed in animals or in cell culture, and are undertaken to predict outcomes in the target clinical population. Nonclinical studies are performed to investigate the factors that affect product performance and quality. These include process and product optimization studies, assay development and validation studies, and process intermediate and final product stability studies. Some studies are performed to provide information that guide company business decisions, whereas other studies are used to help ensure quality and support regulatory requirements.
An evolutionary approach to specification setting should reflect the fact that at early stages of development, manufacturers have limited knowledge of the product and limited clinical experience. During late stage development, there is limited manufacturing information to develop meaningful monitoring tools. Therefore, a staged approach to establishing limits should be considered.
Early-stage development involves predicting relevant quality attributes (and their parameters), methods, and acceptance criteria for the class of molecules based on preclinical characterization, industry experience, published scientific literature, and guidance from regulatory agencies. At early stages of development, selected quality attributes and corresponding limits (acceptance criteria) should be focused on well defined expectations related to product safety (e.g., DNA, Limulus amebocyte assay (LAL), and host cell proteins), and allow clinical experience to define other quality attributes such as product-related substances and impurities. The predicted specifications may be able to take advantage of the knowledge base of other well characterized products and prior research experience.
Specifications are ultimately used to protect the patient from receiving product that is not fit for use. The basis for defining fit for use derives from the preclinical and clinical development experience with drug product, which ultimately is tied to specifications in the product license application. Unless quality attributes can be linked to preclinical and clinical experience, there is no rational basis for establishing specifications. Thus, in early development, measurements of product quality should be made to reliably identify potential quality attributes of the product. The relationship between the level of a quality attribute (such as potency) and its clinical impact is best understood when the attribute is allowed to vary rather than be constrained by narrow limits. This information can then be used to determine which attributes forecast clinical outcomes, and to set limits that ultimately (i.e., during commercial manufacture) predict fitness for use.
Because of this, it is unreasonable to set restrictive specifications for quality attributes that are not directly linked to product safety. Early development data should be unrestricted and thereby scientifically informative. Fit-for-use specifications may be set from these data, and verified in late-stage development.
In late-stage development, selected quality attributes should be based on a biochemical and biophysical understanding of the product that is linked to manufacturing experience and clinical relevance. The selection of analytical methods and corresponding acceptance criteria should be based on sound scientific judgment and an appropriate statistical analysis using knowledge coming from clinical, preclinical, and nonclinical development experiences.
Attributes that are not related to product quality may be either eliminated or reserved for characterization after a manufacturing process change. Such process changes are frequently associated with the preparation for commercialization. Attributes that are redundant with other quality attributes may be eliminated. In principle, the test with the greatest sensitivity and/or reliability should be reserved for conformance testing.
Process performance attributes that are used to monitor consistency rather than fitness for use require adequate experience to reveal the process distribution. This involves a sufficient number of lots manufactured across a representative range of process conditions. Because this is not the primary objective of product development, specifications that are developed to monitor product consistency require data acquired during full-scale manufacture, under normal operating conditions, over a range of changes in process parameters.
Limits related to process consistency may evolve with extended experience in manufacturing. Process control limits should reflect the current state of the process. Thus, if an improvement has been made that has no impact on quality attributes related to safety or efficacy, but results in a shift or change in the distribution of a consistency parameter, process monitoring limits should be amended to reflect the change.
As described in ICH Q6B,1 specifications are the list of quality attributes, methods, and limits. However, for biological and biotechnology-derived products, some quality attributes may have several parameters. For example, purity of a monoclonal antibody (MAb) typically has three parameters related to its size, charge, and molecular integrity. Consequently, specifications for the purity of protein products have four components:
1. Quality attribute (e.g., purity)
2. Parameter (e.g., size, charge, molecular integrity)
3. Analytical method (e.g., size-exclusion chromatography [SEC], ion exchange chromatography, capillary electrophoresis–sodium dodecyl sulfate [CE-SDS])
4. Limit or acceptance criteria (e.g., main peak relative purity for SEC).
It is important to stress that protein products are not structurally homogeneous, such that a single peak on a chromatogram, electropherogram, or in a spectrum may not represent a single molecular entity (as is typically seen with small molecules). No single analytical method can define product purity; therefore, a combination of orthogonal analytical methods typically is used to properly describe a single quality attribute such as purity.
Another quality attribute that is unique to biological and biotechnology-derived products is potency. Potency reflects activity in a biologically relevant system, usually expressed as relative potency to a standard or specific activity. While there is scientific debate regarding the relevance of potency to clinical effectiveness, potency assays typically are considered a link to the mechanism of action of a biological or biotechnology-derived product, which potentially can bridge changes in biophysical characteristics to biological activity.
Specifications can be divided into two categories, product-specific specifications and compendial or regulatory specifications:
Although the focus of most discussions about specifications is about acceptance criteria, a certain clarification about types of specifications, quality attributes, and methods would benefit the industry.
ICH guidelines specify quality attributes for DS and DP, acknowledging that in some cases in-process testing may be more appropriate than DS or DP specifications. Nevertheless, the guidelines specify desirable quality attributes for DS and DP, including the following:
For the most part, the quality attributes for DS and DP are similar. Additional testing specified for DP is associated with changes related to the formulation, vials, or devices. At early stages of development (corresponding to Phase 1–2 clinical trails), many manufacturers do not use the final commercial formulation; instead, they may use a frozen liquid DS or other simple dosage form. Therefore, at this early stage of development, many tests performed on DS and DP may be redundant without adding value for patients or the manufacturer.
Another difference between conformance testing for DS and DP is the treatment of impurities. In testing the DS, the term impurities refers to product and process impurities, whereas in testing the DP, the focus of impurity testing is on degradation products. The purpose of testing for process-related impurities (e.g., DNA, host cell proteins, solvent and buffer components) may not change during drug development. Whether or not expectations regarding impurity testing are well defined in a given regulation, such testing should not be relaxed at any stage of development because impurities are directly related to patient safety. However, robust process validation that demonstrates the removal of impurities (e.g., adventitious agents) may alleviate the need for such stringency. The issue of product-related impurities and product-related substances may merit different considerations. At early stages of development, understanding the presence and chemical structure of product-related impurities may be limited, making a distinction between product-related impurities and product-related substances difficult. Frequently, we refer to them as isoforms, structural variants, or posttranslational modifications. Many of these "modifications" are well known to scientists, but their quantification and physiological significance sometimes remains elusive. In addition, existing literature does not provide unequivocal or uniform evidence about their physiological relevance.3–4 This category includes several well known posttranslational modifications, such as methionine oxidation, deamidation of asparagine residues, and N- and O-glycosylation. Therefore, at the early stage of product development, specifications may not need to focus on product-related impurities. At the later stages, when the understanding of the product increases, and the clinical relevance of product-related impurities and substances can be elucidated, the specifications can be amended to include additional testing for product-related impurities and degradants.
Analytical methods associated with specifications define how a particular attribute will be tested. Therefore, any change in an analytical method may affect the numerical limits and statistical considerations applied in establishing this limit. Because of the continuous progression of analytical technologies, the industry and regulators will need to continuously develop strategies to address the issue of the difference between old and new methods. Modern, more efficient technologies will appear on the market, while old technologies (instruments and consumables) will no longer be supported by current vendors. For example, an evaluation of the history of separation methods suggests that separation technologies have doubled in capabilities in approximately 11 years.5
As analytical methods evolve over time, these new methods frequently provide additional information that was not previously available. In such cases, the relevance of this new information to specifications needs to be evaluated including attributes, methods, and limits.
A hypothetical example follows. At the early stage of product development, an anion-exchange chromatography method was developed to evaluate the charge distribution of a MAb. The chromatogram showed a single main peak and two minor peaks. Based on extensive product characterization and clinical experience, a specification was developed around the relative purity of the main peak, with an understanding that the main peak contains multiple components with the same net charge. The limit of minimum purity was set at 85.0%. At the time of commercialization, the ion-exchange column was discontinued by the vendor, and replaced by the drug manufacturer with a new monolithic ion-exchange column. As a result of this technological advancement, a component of the main peak resolved into two peaks, one of which is a separate small peak accounting for a reduction in relative purity of approximately 5%. In such a situation, adopting old limits for the new method would be inappropriate, and a statistical approach used to create limits for the old method may not be applicable in establishing limits for the new method. The peak resolved by the new method was fully characterized as an isoform containing oxidized methionine in the Fc portion of the antibody, with no effect on potency. In addition, it was demonstrated that the level of oxidation in most recent lots ranged from 4% to 5%. In such a case, the manufacturer should request a revision of the specifications for the main peak without an additional specification for oxidized methionine in the Fc portion. An alternative approach would be to sum up two peaks that were not resolved in the original method.
Chromatographic profiles for biopharmaceutical products present a special challenge. Even at the early stages of development, analytical methods are capable of resolving different isoforms. Understanding their structure–function relationships, however, may require several more years. Therefore, at early stages of development, this limited understanding of isoforms and post-translational modifications may lead scientists to establish specifications exclusively with respect to product purity. In such cases, the levels of product-related forms can be inferred from the relative area of the main peak. The fact that specifications are not designed around product impurities should not prevent the manufacturer from tracking individual peaks (likely containing multiple molecular entities) in another system different from specifications. These data can be part of the characterization data collected during the course of development. The data could be very useful at later stages of development.
The precision of analytical methods is linked to data reporting, whereas specifications limits are linked to method precision.6–9 Therefore, it is very important to use consistent practices regarding the number of significant figures or decimal places reported by analytical methods that is consistent with the number of significant figures or decimal places in the specification limits (acceptance criteria).
The reporting interval (number of decimal places) should be derived from the standard uncertainty, which can be expressed in the form of standard deviation or coefficient of variation (relative std. dev). Published approaches should be evaluated.7,9 The most commonly adopted is the recommendation from the American Society for Testing and Materials (ASTM), which proposes that the results of analytical measurements should be rounded to a decimal place corresponding to not less than 1/20 of the determined standard deviation.
In the biological and biotechnology industries, precision of methods frequently is assessed during qualification and confirmed during method validation.5,10–11 Therefore, appropriate precision studies should be performed before specifications are established.
The purpose of this paper, which has been developed by the Biologics and Biotechnology Working group on specifications of the Pharmaceutical Research and Manufacturers of America (PhRMA), is to provide guidance on a lifecycle approach to setting global specifications for biological and biotechnology-derived products. This Part 1 includes sections 1 to 3 of the paper, covering terminology, the stages of the lifecycle of a product, and components of a biological and biotechnology product specification. Parts 2 and 3 of this article will be published in the next two issues of BioPharm International. Those parts will include section 4, covering current issues related to the development of specifications; and section 5, which suggests an approach for developing and maintaining a total quality system.
Izydor Apostol, PhD, is scientific director, analytical and formulation sciences, at Amgen Inc., Thousand Oaks, CA; Timothy Schofield is senior director, nonclinical statistics, at Merck Research Laboratories, Merck & Co., West Point, PA, and the corresponding author, email@example.com 215.652.6801; Gerhard Koeller, PhD, is vice president, quality and compliance biopharmaceuticals, at Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany; Susan Powers, PhD, is biotech quality technology leader, Wyeth Pharmaceuticals, Collegeville, PA; Mary Stawicki is associate director, regulatory affairs biopharm CMC, at GlaxoSmithKline, Collegeville, PA, and Richard A. Wolfe, PhD, is director and team leader, biopharma operations, at Pfizer Global Manufacturing, Chesterfield, MO.
1. International Conference on Harmonization. Q6b, Specifications: Test procedures and acceptance criteria for biotechnological/biological products. Geneva, Switzerland; 1999.
2. Swarbrick J, Boylan JC. Encyclopedia of pharmaceutical technology. New York: NetLibrary, Inc.; 2001.
3. Koren E, Zuckerman LA, Mire-Sluis AR. Immune response to therapeutic proteins in human—clinical significance, assessment and prediction. Curr Pharm Biotechnol. 2002;3:349–360.
4. Graves DJ, Martin BL, Wang JH. Co- and post-translational modification of proteins. Chemical principles and biological effects New York: Oxford University Press; 1994.
5. Apostol I, Kelner D. Managing the analytical lifecycle for biotechnology products: A journey from method development to validation. BioProcess Int. 2008; in press.
6. Horwitz W, Albert R. A heuristic derivation of the Horwitz curve. Anal Chem. 1997;69: 789–790.
7. ASTM. Standard Practice for using significant digits in test data to determine conformance with specifications. In ASTM Designation: E 29–04 ed. 2005; p 1–5.
8. Holme DM, Peck H. Analytical Biochemistry. 3rd ed. Upper Saddle River, New Jersey: Prentice-Hall, Inc.; 1998, p 9–19.
9. Agut C, Segalini A, Bauer M, Boccardi G. Relationship between HPLC precision and number of significant figures when reporting impurities and when setting specifications, J Pharm Biomed Anal. 2006;41:442–448.
10. International Conference on Harmonization. Q2R1, Validation of analytical procedures. Geneva, Switzerland; 2005, p 1–13.
11. Ritter NM, Advant SJ, Hennessey J, Simmerman H, McEntire J, Mire-Sluis A, Joneckis C. What Is test method qualification? Proceedings of the WCBP CMC strategy forum, 24 July 2003. BioProcess Int. 2004;2,32–46.